Annexes Civic Participation and Political Participation Data availability issues and imputations Indicators in these two domains remain unchanged from the 2013 YDI. But there is an improvement in the quality and coverage of the data in the 2016 YDI. The methodology has been designed to be in line with other prominent global indicators, and substantial effort has been made to populate the index with the best existing country information. However, the major challenge to developing a harmonised youth development index is in attempting to overcome the paucity of consistent and comprehensive data coverage across countries varying significantly in land mass, population, level of economic development and location. Data difficulties are particularly acute for civic and political indicators where the best available attitudinal data have been selected. One of the major outputs of this process is a summary not only of the available data, but also of the data that cannot be currently sourced from the existing stock of data. A2.3  Overview of data limitations There were many data limitations to overcome in developing the 2016 YDI. Many of the same challenges are likely to be encountered in the production of national or regional YDIs. Data limitations include: • data not being available on all 18 indicators for all the countries in the YDI • some data being for the 15–24 age group and others for the 15–29 age group • finding representative data on political participation and civic participation – the greatest challenge when seeking data on the domains in the YDI • national averages hiding regional variations, which could mask serious inequalities in a country with a large population or land mass • the impossibility of disaggregating some data in the YDI by gender. The issue of low availability for current or historical data has been a factor in a number of the methodological decisions made, from what indicators to include to how to calculate the final scores. Many empirical and statistical techniques can be employed to deal with the challenge of finding missing data when creating a composite index.4 Table A2.6 lists them and explains how these applied, or did not apply, to the YDI. Table A2.6  Data imputation methods applied in the YDI, in order of application Imputation method Description Application in YDI Time series imputation Replace missing values using linear interpolation When at least two data points exist in time for an indicator-country pair, linear interpolation is used to estimate data for unreported years. Cold deck imputation Replace the missing value with a value from another source When only one data point exists for an indicatorcountry pair, these data are extrapolated for all years. Hot deck imputation Assign missing data the value of a ‘similar’ data point Where time series and cold deck imputations fail, indicator-country pairs are assigned averages of other countries in the same year in the following order of preference: 1. the average of the country’s region 2. the average of other countries in the same income bracket as the country as defined by the World Bank 3. the average of all other countries with the same government type as the country as defined by the Economist Intelligence Unit 4. the global average. Only the most preferable of the four hot deck imputation techniques listed is used for any single missing data instance. \ 141